[Syzygies]

Here is a design for a 36 step "volume, trim" stepped attenuator small enough to fit inside a Hammond 1455J12 case. Each rotary switch is 1/2" in diameter. I've spread out the parts for clarity, but they could easily be wired more tightly:

The circuit is a nested pair of "series" type stepped attenuators. See How Stepped Attenuators Work for background. Volume, trim attenuators are common, and were discussed in the thread "Bicycle Shifting" Stepped Attenuators?. There, no one was able to dig up an existing schematic. This is the simplest circuit I can think of that might work. I'll be satisfied if it sounds better than the pots that fit in this small a space. The Alps Blue doesn't, and single-knob stepped attenuators tend to be bigger than the amps I want to build.

It is important to wire the trim control before the volume control, to minimize the effects of current division, and to keep the varying input impedance in a tighter range. My idea is to solder 0805 surface mount resistors directly to the switch solder lugs; they fit perfectly, and this avoids etching a circuit board. I asked for advice on choosing surface mount resistors in the thread Best surface mount resistors for stepped attenuator?. I'm thinking of 0.1% values because cutting resistors to this accuracy is said to reduce noise.

If one could live with 12 volume levels, a simple variation would be to build separate series or shunt stepped attenuators for each channel, same form factor and a similar parts list.

One could perhaps get away with using non-shorting switches by adding four more resistors and carefully adjusting all values. However, NewarkInOne will special order in single quantities what looks like EAO's exact replacement for Grayhill's shorting 56SD30-01-2-AJS rotary switch, on which this design is based. I cannot get specs from EAO's site, so it looks like I will have to order two and hope for the best.

http://www.newark.com/NewarkWebComme...sp?SKU=70C1794
http://embrace.grayhill.com/embrace/...DF/J-22-25.pdf
70C1794
56SD30-01-2-AJS
2 x $16.89 = $33.78

http://www.mouser.com/index.cfm?&han...*&N=0&crc=true
Vishay/Dale TNPW Series 1/10W 0805 Precision Thin Film Chip Resistors 0.1%
20 x $0.85 each = $17.00

http://www.mouser.com/index.cfm?&han...*&N=0&crc=true
Vishay/Dale MODEL CMF, RN55D 1/8W 1% Metal Film Resistors
0.240x.090in, .025in leads
4 x $0.16 each = $0.64

 

[doobooloo]

Wonderful idea. Please keep us updated regarding your progress!

But I'm curious - is there any amp so portable that is good enough to really deserve such an elaborate volume control system?

 

[Syzygies]

Quote:

Originally Posted by doobooloo But I'm curious - is there any amp so portable that is good enough to really deserve such an elaborate volume control system?

i've just ordered parts for a couple of MINTS and a couple of PIMETAS. I'll build one of each by the book, and play with the other. So I'll find out!

From my readings, I have the following impressions:

1. A PIMETA can sound pretty good.
2. Any pot that can fit in the 70 x 24 mm space inside a Hammond 1455J12 case sounds crummy and tracks worse.
3. Stepped attenuators do sound distinctly better than pots.

The J12 case is very portable. If I can solve the power and pot problems, I should end up with a very high end amp for the form factor.

The other solution I've been considering is a 12 step garden variety series attenuator, but with extra resistors to allow the use of a non-shorting switch. I'd like to use

Grayhill 71BDF30-02-1-AJN SWITCH ROTARY 1P-12POS 2 DECK
http://www.digikey.com/scripts/DkSea...107-ND&Site=US
GH7107-ND $30.62
http://www.newark.com/NewarkWebComme...sp?SKU=70B7173
70B7173 $25.39 (trust the part # not the description?)

It also fits in the J12 case (I had the scale wrong before; everything has more room than I had shown):

The circuit uses the same 0805 surface mount resistors. For both of these circuits, finding resistor values involves solving quadratic equations, easy, but messier than ordinary series attenuators:

 

[Syzygies]

Chiefdom Rotary Switches were discussed as an aside today on the thread Interest check: DACT CT2 2 Deck Stepped Attenuator.

For my purposes this is the most exciting switch I've seen so far. For small 12 step attenuators, I like the specs of their RCP 203 series, available 2 and 4 decks, shorting. By-the-book ladder and series attenuators lay out perfectly using 0805 resistors. I'm going to contact them to see what the deal is, see if I can buy in sample quantities.

A series circuit:

A ladder circuit:

The ladder circuit, cased:

 

[Syzygies]

I spoke with Grayhill and then with a distributor. Their 56A30-01-2-06S switch (for my first design in this thread) can be ordered, with an indeterminate lead time:

56 Series
A Standard 1/8" shaft w/ solder lugs
30 Degrees per step
-01 Decks
-2 Poles per deck
-06 Thows per pole
S Shorting

1-24 $15.00
25-99 $11.95
100+ $10.75

Most of their switches are build-to-order, including pretty much any shorting switch. They like getting large orders through distributors, but the distributor can bundle lots of 1's and 2's custom built as part of such an order. Distributors that normally deal retail, like the usual suspects we know, are least likely to do this for us. The smaller distributors who deal primarily with companies are more likely to treat one of us like we're an engineer, and build samples for us paid by credit card. It doesn't hurt to call anything you're doing a prototype (it is!) and mention that our community does lots of group buys.

I somewhat struck out with build-to-order shorting Grayhill 71 series switches for series or ladder attenuators. They'll make 71ADF30-02-1-AJS or 71AF30-02-1-12S-C but not 71AF30-02-1-12S-F or 71AF30-04-1-12S-F.

English to english translation: If I want a stop between max and min volume on their 71 series, I have to buy an adjustable "prototype" switch, and 4 decks is no go, I'd have to stick to a 2 deck series attenuator design.

 

[doobooloo]

Are you planning to get the resistor network boards professionally made, or just hand-solder the resistors to the switches?

If you could somehow get a group buy for those Chiefdom switches going as well as one for accompanying ladder/series resistor boards then I think quite a few people may be interested.

 

[Syzygies]

I've written Chiefdom for samples. I want to hand-etch and hand-solder several prototypes. Since these are textbook circuits, not one of my funny circuits, they should work, it's just a question of how well. (Well, my ladder layout is a bit different, but I'm guessing that the difference doesn't matter.)

If I build something that makes me happy, I'll offer loaner prototypes to any golden ears who want to listen. If there's any consensus at that point this is worth doing, I'm happy to organize the most sensible group buy we can think up. I get the impression that my circuit could easily be machine built, but buying boards would be cheaper.

What I like about home etching is the ability to make something after you've missed the group buy. I still want to work out plans for a home-etched diamond buffer. I wish I could find a ready source for single-quantity shorting switches like the Chiefdom, so this could be a home-etch project.

Edit: Grayhill did say over the phone that they make far more of the 71ADF than the 71BDF (A = 1/8" shank, B = 1/4" shank). The easiest to locate Grayhill part number for a textbook 12 step series attenuator like I've drawn is:

71ADF30-02-1-AJS

(The Chiefdom drawing needs a slight modification to fit the Grayhill pinout.)

Newark lists it with a long lead time for $25.86:
http://www.newark.com/NewarkWebComme...sp?SKU=39C3100

AVNet has better prices, with a twist on the part number to search for (actually the same part):
http://www.em.avnet.com/home/

Series, 2 decks:
71ADF30-02PAJS
1 $21.76
25 $17.41
100+ $13.01

Ladder, 4 decks:
71ADF30-04PAJS
1 $33.57
25 $26.856
100+ $20.142

With a decent resistor for less $$$ than I spec'd, these prices get very appealing.

 

[Syzygies]

I played around tonight with a couple of ladder layouts for the Grayhill 71 series. I must say, once one gets the hang of Adobe Illustrator, I think it's much faster to use than the dedicated circuit software I've seen, for sketching circuits. I plan to home-etch; of course one needs to use circuit software to prepare professional boards, and I admire people who find that easy.

This uses two 71ADF30-02QAJS 4x6 switches, for a 36 step "volume, trim" attenuator, two nested ladder attenuators:

This uses one 71ADF30-04PAJS 4x12 switch, for a 12 step standard ladder attenuator:

In both circuits, the other side of the PCB is a ground plane, with a few throughs to hook up ground to the switches and input, output.

Avnet, Inc. is the best source for either of these shorting switches, with a 6 week lead time:

71ADF30-02QAJS
1, $22.9120
25, $18.3300
100, $13.7470

71ADF30-04PAJS
1, $33.5700
25, $26.8560
100, $20.1420

These attenuators use a lot of resistors, though less than typical high-end attenuators. I think I'll wait until the PPA v2 is out to consider using one; I'm going back to the smaller 56 series for my Pimetas.

 

[Syzygies]

Here are two circuit board layouts for a 36 step "Volume, Trim" series-type stepped attenuator, using a pair of 1/2" Grayhill Series 56 rotary switches:

2 x Grayhill 56P30-01-2-06S 2x6 switch
www.avnet.com
1, $12.2970
25, $9.8380
100, $7.3780

24 x 1206 surface mount resistors, e.g.
Vishay TNPW series, $0.69 each, 100 @ $0.48 each.

So parts cost for one attenuator, not including the board, is a bit over $40, a bit over $30 each for 25.

Again, the other side is a ground plane, with a few throughs to reach ground. This needs two short jumpers to carry signals between the switches; one could also squeeze the connections onto the ground plane side.

 

[Syzygies]

Here's a first draft of a through-hole 36 step attenuator like the preceding plan, only a bit bigger:

With decent $0.16 resistors like the Dale Vishay RN55's, this brings the resistor cost down to $4 or so.

 

[doobooloo]

Ooh. I really like the through hole version.

Keep us updated!

Amazing work!

 

[Syzygies]

Here's a tighter through-hole version, and the analysis with resistor values. It again uses 24 resistors for a 36 step attenuator.

Since so many commercial attenuators are complex 3D contraptions, I decided it was ok to use two boards, one to reposition the switch pins to where I want them. One can see that I borrowed a bit from the DIP 16 step attenuator design. I may use his socket idea to easily experiment with resistor values, although for a final version I don't want the extra mechanical contacts.

Comparing designs, one can see that the surface mount version is much simpler to etch and drill, and has a shorter signal path. There would be a huge payoff to getting the hang of surface mount soldering. On the other hand, the final result here is a similar size, and good leaded resistors are cheaper. Glassman reports many people had problems soldering his surface mount buffer kit. I view this version as a prototype for ordering professional boards; this design is simple if someone hands you the boards:

Here are all layers, cased:

Here are the traces for the single-sided front switch board:

Here are the traces for the double-sided back resistor board, both sides shown superimposed. Use resistor leads where possible to connect the two boards; four additional center pins are needed for switch inputs. The trim is on the right where the inputs enter; the volume is on the left where the outputs exit:

Here is the circuit schematic:

Here are some reasonable choices for resistor values, with some analysis of the circuit for these choices:

The spreadsheet picks closest available resistors from a list. I've posted it as VolTrim36.xls.zip in case anyone wants to lift the code for other purposes.

Here is the key Mathematica formula for picking the first stage resistor values:

I can only guess how to best set these resistor values. When we read for an ordinary pot "too low an impedance is bad for the source device", I'm thinking combined effective impedance of both stages here. When we read for an ordinary pot "too high an impedance is noisy", I'm thinking the impedance of either stage considered separately. I consider it particularly important to keep the input impedance range tight, so one isn't suffering a "phases of the moon" change in sound quality as one fiddles with the trim.

As I said before, it is crucial to put the trim stage first; these numbers come out much worse the other way around.

 

[Syzygies]

This design is for a 36 step "Volume, Trim" stepped attenuator using through hole resistors parallel to the switch body, similar e.g. to the Goldpoint attenuator pictured in http://207.228.241.188/mv_am_3.jpg. The board is single-sided. This looks easier to make than any of my other designs. There is room for Vishay RN55D, Welwyn RC55Y, or similar resistors.

2 x Grayhill 56P30-01-2-06S 2x6 switch
www.avnet.com
1, $12.2970
25, $9.8380
100, $7.3780

24 x Vishay RN55D Resistors
Mouser, $0.16 each, 50 @ $0.11 each.

So parts cost for one attenuator, not including the board, is under $30. For comparison, 24 step Goldpoint kits start at $150, and don't fit the Hammond 1455J12 case.

 

[Syzygies]

I heard back from Chiefdom Rotary Switches:

Quote:

Actually our min. q'ty is 1000pcs per item, but, if you request on hundreds then, you can take goods via Courier way with freight collect. You didn't mention what's shaft type and shaft length you need, so, pls kindly advise I can send you quotation to you first as follows; RCP203-2-2-12SA ( 2 steps, 2 pole, 12 position, shorting with stop at 12 position ) PRICE @USD 3.00 ( WITH SILVER-PLATED ) @USD 3.05 ( WITH GOLD-PLATED ) RCP203-4-4-12SA ( 4 steps, 4 pole, 12 position, shorting with stop at 12 position ) PRICE @USD 4.00 ( WITH SILVER-PLATED ) @USD 4.30 ( WITH GOLD-PLATED ) MIN. Q'TY: 1000PCS PAYMENT: BY T/T ONCE ORDER CONFIRMATION DELIVERY: WITHIN 35 DAYS AFTER RECEIPT OF YOUR T/T

To recap, one could etch a 12 position ladder attenuator for the RCP203-4-4-12SA using 0805 surface mount resistors, such as ??

http://www.mouser.com/index.cfm?&han...*&N=0&crc=true
Vishay/Dale 0805 CRCW Series Thick Film Chip Resistors
48 x $0.11 each = $5.28

using the circuit layout

This is $10 for a 12 step ladder stepped attenuator, $20 to nest two in "Volume, Trim" fashion for 144 steps. I'd want to go with 0.1% resistors while I was at it, but I'm curious if these resistors already out-perform ordinary pots in this price range? In any case they'd make great cheap soldering practice...

I had pretty much settled on ordering a handful of Grayhill 56's, and testing my 1206 surface mount design:

I

However, I should take Chiefdom up on their sample offer?

 

[doobooloo]

Wow! more designs!

Out of curiosity, how do you create your graphics? They look very professional.

Quote:

However, I should take Chiefdom up on their sample offer?

Can you try both the Chiefdom and the Grayhill?